# Gemini CLI Gemini CLI is an open-source agentic coding CLI tool created by [[Google]]. It is similar to [[Claude Code]] and [[OpenAI Codex]]. It uses a ReAct (reason and act) loop with built-in tools and [[Model Context Protocol (MCP)]] servers to handle complex development tasks like fixing bugs, creating features, and improving test coverage. > [!warning] Being replaced by Antigravity CLI > Announced 19 May 2026 (by Dmitry Lyalin and Taylor Mullen on the Google Developers Blog): Google is consolidating its developer tooling onto the [[Google Antigravity]] platform, and [[Antigravity CLI]] is the successor. Gemini CLI and Gemini Code Assist IDE extensions stop serving consumer users (Google AI Pro / Ultra, and the free [[Gemini Code Assist]] for individuals) on **18 June 2026**. **No new GitHub organization installations** are permitted from the same date; existing GitHub-triggered requests stop shortly after. Enterprise Gemini Code Assist Standard / Enterprise licences and direct API-key users keep it indefinitely. Migration guide: https://antigravity.google/docs/gcli-migration Gemini CLI is also the engine behind [[Gemini Code Assist]]'s agent mode in [[Visual Studio Code (VSCode)]]. ![[Gemini CLI Image.png]] ## Key Features - Built-in tools: grep, terminal, file read/write, web search, web fetch - [[Model Context Protocol (MCP)]] support for third-party tool integration - [[AI Agent Skills]] support - Image input (screenshots, diagrams) - Memory management via `/memory` command - Non-interactive (headless) mode for scripting and automation - Pre-installed in Google Cloud Shell ## Commands - `/memory`: manage persistent context - `/stats`: usage statistics - `/tools`: list available tools - `/mcp`: MCP server configuration - `!`: toggle shell mode ## License and Privacy Gemini CLI itself is open source (Apache License), but its usage implies acceptance of terms and conditions. When people use their personal Google account, the "Gemini Code Assist Privacy Notice for Individuals" applies. Under that notice, prompts, answers and related code are collected and may be used to train Google's models. When people use a Gemini API key, Gemini API terms of service apply, which also means that prompts, answers and code are collected and may be used for training AI models. The only case where prompts, answers and code are not used for training is when users pay for Gemini either via Google Cloud or the Gemini API. In that case inputs are considered confidential. But apparently, NOT the outputs! ## Tips & Tricks ### Google search alias Use Gemini CLI's web search tool as a quick `google` command from the terminal: ```sh google(){ gemini -p "Search google for <query>$1</query> and summarize the results" } ``` Usage: `google "latest claude code features"` ## References - Documentation: https://developers.google.com/gemini-code-assist/docs/gemini-cli - Source code: https://github.com/google-gemini/gemini-cli - Website: https://geminicli.com - X account: https://x.com/geminicli - Created by Taylor Mullen: https://x.com/ntaylormullen - Transition announcement (19 May 2026, Dmitry Lyalin + Taylor Mullen): https://developers.googleblog.com/an-important-update-transitioning-gemini-cli-to-antigravity-cli/ - Migration guide: https://antigravity.google/docs/gcli-migration - Analysis by [[Simon Willison]]: https://simonwillison.net/2025/Jun/25/gemini-cli - System prompt: https://gist.github.com/simonw/9e5f13665b3112cea00035df7da696c6 ## Related - [[AI Agent Harness]] - [[Google]] - [[Google Antigravity]] - [[Antigravity CLI]] - [[Gemini Code Assist]] - [[Claude Code]] - [[OpenAI Codex]] - [[Kimi CLI]] - [[GitHub Copilot CLI]] - [[AI Agents]] - [[AI Agent Skills]] - [[Model Context Protocol (MCP)]] - [[Agentic AI Foundation (AAIF)]]